Menu

#36 N_GLOBAL_NODES overflow

v1.0 (example)
open
nobody
None
5
2019-05-29
2019-05-24
Laurent
No

The N_GLOBAL_NODES overflow error is not documented.
Through the C code, I think that the maximum of subckt nodes is 1005 : is that the wright understanding ?

BRgds,
Laurent

Related

Support Requests: #36

Discussion

  • marcel hendrix

    marcel hendrix - 2019-05-24

    On 2019-05-24 11:35, Laurent wrote:


    ** [support-requests:#36] N_GLOBAL_NODES overflow**

    Status: open
    Group: v1.0 (example)
    Created: Fri May 24, 2019 09:35 AM UTC by Laurent
    Last Updated: Fri May 24, 2019 09:35 AM UTC
    Owner: nobody

    The N_GLOBAL_NODES overflow error is not documented.
    Through the C code, I think that the maximum of subckt nodes is 1005 :
    is that the wright understanding ?

    BRgds,
    Laurent

    Is this one or two questions?
    The biggest circuit I have here has more than a million nodes.
    When I wrap it in a single subckt it still works?

    Number of lines in the deck = 1603640
    Netlist loading time = 5.670 sec.
    Netlist parsing time = 0.265 sec.
    

    -marcel

     

    Related

    Support Requests: #36

  • Laurent

    Laurent - 2019-05-24

    This a remark about the documentation of : "N_GLOBAL_NODES overflow"

    Second, I found out that it is a limitation in the number of pin of a subckt : < 1005
    Is it right that the number of subckt pins is limited ?

    Laurent

     
  • Dietmar Warning

    Dietmar Warning - 2019-05-24

    Yes, it is right.
    Extract from subckt.c:

    define N_GLOBAL_NODES 1005

    /---------------------------------------------------------------------
    * table is used in settrans and gettrans -- it holds the netnames used
    * in the .subckt definition (t_old), and in the subcircuit invocation
    * (t_new)
    --------------------------------------------------------------------/
    static struct tab {
    char
    t_old;
    char t_new;
    } table[N_GLOBAL_NODES]; /
    That had better be enough. */

    Maybe the naming is a bit misleading for the error message.

    Dietmar

     
  • Laurent

    Laurent - 2019-05-29

    I suggest to place N_GLOBAL_NODES as an option.
    I agre that 1005 shoud be OK for most of the circuit.
    In my case, I want to simulate a very large array of cells (1000x1000), with n inputs/outputs so the connection to the array is : 1000 x n I/Os
    If I could change that value with an option N_GLOBAL_NODES, NGspice would be perfect !

    Thanks! Best regards !
    Laurent

     

Log in to post a comment.